Phantom Activation Volumes - American Chemical Society

This gives rise to the illusion of volume shrinkage along the reaction coordinate toward the transition state, which we term phantom ΔV*. Thus the tr...
0 downloads 0 Views 502KB Size
J. Phys. Chem. A 2000, 104, 3057-3063

3057

Phantom Activation Volumes1 Kevin A. Swiss and Raymond A. Firestone* Department of Medicinal Chemistry, Boehringer Ingelheim Pharmaceuticals, Inc., 900 Ridgebury Road, Ridgefield, Connecticut 06877-0368. ReceiVed: October 7, 1999; In Final Form: January 20, 2000

Activation volumes ∆V* are presently obtained by measuring the effect of pressure on the reaction rate. It is implicitly assumed that the entire response of rate to pressure is volume related, i.e., acceleration by high pressure reveals shrinkage as the reactants progress to the transition state, and vice versa. However, we now demonstrate that high pressure accelerates some bond-making reactions in an additional, nonvolume-related way, through its elevation of solvent viscosity. Diels-Alder reactions, 1,3-dipolar cycloadditions, and Claisen rearrangements are accelerated by rising viscosity and are therefore subject to viscosity-associated acceleration at raised pressures. This gives rise to the illusion of volume shrinkage along the reaction coordinate toward the transition state, which we term phantom ∆V*. Thus the true ∆V* for these reactions, while negative, is less negative than previously believed. Corrections in ∆V*, calculated from experimental rate-viscosity plots, range up to 61%.

Introduction Volume of activation (∆V*) is an important criterion of mechanism. It is defined in eq 1.2

∆V* ) -RTδ logk/δP

(1)

Acceleration of a reaction by pressure means that the size of the transition state (TS), which includes not only the reacting atoms but also the surrounding solvent molecules, is smaller than that of the reactants, and retardation by pressure means that the TS is larger. For example, bond-forming reactions are usually pressure accelerated and bond-breaking reactions pressure retarded. However, sometimes solvent effects are overwhelming, as with solvolyses which, although bond breaking, are pressure accelerated owing to electrostriction of solvent by the polar TS.3 Another example of solvent domination is the Diels-Alder reaction. It is pressure accelerated as expected for a bond-forming reaction, but the ∆V* term arises predominantly from reduction of empty space in the solvent around the TS rather than from bond formation itself.4 It is generally assumed that the ∆V* calculated from eq 1 is truly a volume term. It is our thesis, however, that ∆V* is only partially volume derived, because increasing the pressure can induce kinetic effects that do not arise from volume changes. We propose the term “phantom activation volumes” for pressure-induced rate changes that are defined as ∆V* in eq 1 but are in fact not volume related. It is important to identify them because otherwise erroneous decisions regarding mechanism might be made. How can phantom ∆V*s come about? One way is to have a reaction whose rate is sensitive to solvent polarity. Since solvent polarities rise modestly with pressure,5 such a reaction would respond to increased pressure by a rate increase caused, not by shrinkage in the TS, but instead by the pressure-induced increase in solvent polarity.6 Of course, superimposed on this rate * Corresponding author. E-mail: [email protected]. com. Fax: 203-791-6072.

increase might be another arising from electrostriction (a true volume effect), and yet another positive or negative rate effect resulting from bond formation or breakage. Thus only a portion of the ∆V* calculated from eq 1 would be a true ∆V*, and the other portion a phantom (negative) ∆V*. Another possible source of phantom ∆V*s is the fact that solvent viscosity usually rises geometrically with pressure.7 The isomerization of azo compounds and benzylideneaniline, which is normally slowed by high viscosity, shows a strong pressureinduced retardation that arises, not from shrinkage, but instead from the increase in viscosity created by the pressure.8 In this instance the viscosity-derived phantom ∆V* is positive, but a negative one can also come about (vide infra).1,9,10 The Diels-Alder Reaction Diels-Alder (DA) reactions are strongly accelerated by pressure, with ∆V*s ranging from -30 to -50 cm3/mol, which are not far from the overall reaction volumes, ∆VR (∆VR ) Vproduct - Vreactant).11 This has been interpreted in terms of a concerted mechanism, in which both new bonds are forming in the TS, rather than a stepwise-diradical mechanism, where only one new bond is created in the rate-determining step (RDS). We have opposed this interpretation4 on two grounds: (1) a concerted mechanism would have an early TS while a diradical one would have a late RDS TS, so that contraction arising from bond formation would not differ much between them; (2) shrinkage at the TS is so dominated by reduction in empty space among solvent molecules that bond formation hardly affects ∆V* at all. Thus drawing mechanistic conclusions is not warranted. Another controversy involves the observation that |∆V*| often equals or even exceeds |∆VR|,11 i.e., the TS is apparently no larger, and may even be smaller, than the cycloadduct. This has been attributed to secondary orbital interactions, but they can be ruled out since often endo and exo cycloadditions do not differ in their ∆V*s.12 It is impossible for a partially formed bond to be the same size, not to mention smaller than a fully formed bond.

10.1021/jp9935900 CCC: $19.00 © 2000 American Chemical Society Published on Web 03/11/2000

3058 J. Phys. Chem. A, Vol. 104, No. 13, 2000

Swiss and Firestone

We propose that this apparent paradox can be resolved by recognizing that a significant portion of the published negative ∆V*s for DA reactions are phantom volumes. This is because solvent viscosities rise geometrically with increasing pressure7 and DA rates rise with increasing viscosity.13 This was first recognized more than a decade ago,10 in an intramolecular case (reaction 1) run in mono-, di-, tri-, and tetraglyme. Unfortunately, quantitative calculation of phantom volumes was not then possible owing to our somewhat imprecise viscometry, and also to the fact that viscosity data had been obtained on pure solvents but not the reaction mixtures themselves.

We now report accurate viscosities of the actual reaction mixtures,1 enabling the determination of a reliable figure for the rate-viscosity relationship. This relationship is also reported for another DA, the dimerization of cyclopentadiene (reaction 2), run in pure linear saturated hydrocarbon solvents.1 For this intermolecular case, as viscosity rises to about 1-2 centipoise (cP), the rate first rises and then falls. The rise stems from viscosity-induced acceleration in the collision-controlled regime. Only the rising portion of the curve concerns us here, because it is over this viscosity range that published activation volumes were determined. These rate-viscosity relationships permit the determination of what the accelerations would have been in the pressure-accelerated DA cases had they arisen solely from viscosity effects. The viscosity-induced acceleration turns out to be a significant factor of the observed pressure-induced acceleration. Clearly, therefore, all published ∆V*s are too negatiVe for DA and other reactions that are viscosity-accelerated, e.g., Claisen rearrangements (vide infra).

The simplest relationship between rate and viscosity is a linear one, which was permitted by the 1981 data and tentatively adopted at that time for parsimony’s sake.10 However, after examination of all the DA pressure-rate data11 for which the solvents’ pressure-viscosity relationships were available,7 it became apparent that only a log-log relationship (natural log, base e) gives straight lines (Figures 1 and 2).14 We therefore calculated the phantom volumes in this manner. Since ∆V*s are usually based on the initial slopes, i.e., at atmospheric pressure,2 the calculations involve little if any extrapolation beyond our range of measured viscosities. Figures 1 and 2 depict typical cases from the literature of rates vs viscosities. The viscosities were obtained by replacing the pressures in the original kinetic papers by the corresponding viscosities at those pressures for those solvents. 7 Of course only a portion of the slopes arises from viscosity effects, but the linearity is very good up to ∼1 - ∼2 cP,15 and this is significant because the viscosity-induced rate effects are a large portion of the total. Table 1 provides literature data for a large selection of intermolecular DA cases for which both apparent ∆V*s and solvent viscosities under pressure are available, along with the phantom ∆V*s. These are defined as that portion of the apparent ∆V*s that arise from viscosity rather than from true volume

Figure 1. Log reduced viscosity vs log reduced rate of selected literature data. Data taken from Table 1 entry 4 (symbol 1), entry 5 (symbol b), entry 17 (symbol O).

Figure 2. Log reduced viscosity vs log reduced rate of selected literature data. Data taken from Table 1 entry 10 (symbol O), entry 13 (symbol b), entry 19 (symbol 1).

effects. Viscosity effects on the rates were taken as the initial slope (0.96) of log reduced rate vs log reduced viscosity for reaction 2, and of necessity were all assigned the same slope, clearly an approximation. Phantom activation volumes, then, are the ratios of 0.96 and each total slope, times apparent ∆V*s. In all entries the closest temperature between reaction and literature pressure-viscosity data was used (e 20 °C); extrapolation or interpolation with respect to temperature was not done. Phantom ∆V*s are as high as 61% and true ∆V*s as small as 39% of the apparent ∆V*s, a significant correction (but vide infra). Table 2 provides three literature examples of intramolecular DA cases for which both apparent ∆V*s and solvent viscosities under pressure are available, along with the phantom ∆V*s which were calculated as before, but using the slope (0.204) of the intramolecular DA (reaction 1) in ref 1. In all entries the closest temperature between reaction and literature pressureviscosity data was used. In entries 2 and 3 which were done at 153 °C in hexane, we were limited to use pressure-viscosity data of 75 °C.16 Phantom ∆V*s are as high as 21%, and true ∆V*s as small as 79% of the apparent ∆V*s, a significant correction (but vide infra). We recognize, of course, that the use of only a single case each of intermolecular and intramolecular DA to calculate phantom activation volumes for all cases is only semiquantitatively justified, since rate-viscosity slopes undoubtedly vary from case to case. Thus only approximate conclusions can be drawn, except for CPD itself.

Phantom Activation Volumes TABLE 1: Diels-Alder Cycloadditions

J. Phys. Chem. A, Vol. 104, No. 13, 2000 3059

3060 J. Phys. Chem. A, Vol. 104, No. 13, 2000

Swiss and Firestone

TABLE 1: Diels-Alder Cycloadditions

a Pressure-viscosity data of 25 °C was used (ref 7c). b Pressure-viscosity data of 30 °C was used (ref 7a). c Pressure-viscosity data of 60 °C was used (ref 7c). d Pressure-viscosity data of 75 °C was used (ref 7a). e Pressure-viscosity data of 40 °C was used (ref 7c).

TABLE 2: Intramolecular Diels-Alder Reactions

a

Pressure-viscosity data of 30 °C was used (ref 7a). b Pressure-viscosity data of 75 °C was used (ref 7a).

Although it is clear that phantom ∆V*s are indeed large, the corrections do not settle the mechanistic controversy of concerted vs diradical, but they do resolve the paradox of TSs seeming impossibly small. In all the cases where (|∆VR| |∆V*|) appeared to be